Term project on “Flood Frequency Analysis” for “Water Resources in Changing Environment” class at University of Central Florida.
This interactive web application was built using the Shiny package in R for showing the results of flood frequency analysis.
U.S. Geological Survey’s (USGS) National Water Information System.
Peak discharge data for 8 gauge stations on Pearl river, Mississippi were selected for the flood frequency analysis.
Figure 1: Map of gauge stations locations on Pearl river
| Station | USGS Code |
|---|---|
| Jackson | USGS02486000 |
| Edinburg | USGS02482000 |
| Carthage | USGS02482550 |
| Lena | USGS02483500 |
| Rockport | USGS02488000 |
| Monticello | USGS02488500 |
| Columbia | USGS02489000 |
| Bogalusa | USGS02489500 |
We used following probability distributions for modelling annual maximum streamflow time series:
Normal distribution
Lognormal distribution
Gamma distribution
Pearson type 3 distribution
Log-Pearson type 3 distribution
Gumbel distribution
Weibull distribution
Exponential distribution
We selected following methods for estimating parameters of the distributions for this study:
Maximum Likelihood Estimation (MLE)
Method of Moments (MOM)
Probability Weighted Moments (PWM)
Goodness-of-fit tests are used to summarize the discrepancy between a statistical model and the observed data. They are useful for comparing the observed values with either the values fitted by a model of interest or theoretical quantiles of a known sampling distribution. We used following metrics for determining whether a fit is satisfactory or not.
Root-Mean-Square Error (RMSE)
Kolmogorov-Smirnov test (K-S)
Anderson-Darling test (A-D)
Akaike Information Criterion (AIC)
Bayesian Information Criterion (BIC)
L-moment ratio diagram
Javed Ali
PhD Student in Civil Engineering (Major: Water Resources Engineering)
Department of Civil, Environmental and Construction Engineering
National Center for Integrated Coastal Research (UCF Coastal)
University of Central Florida
Email: javedali@knights.ucf.edu